A Multi-AUV Bearings-Only Multi-target Tracking Method Based on the Fast LMB Filter

Autonomous underwater vehicles (AUVs) with bearing target tracking capability are fundamental and essential. However, they face the problems of high nonlinearity, difficult target trajectory initialization, and poor multi-target tracking (MTT) performance. Consequently, we adopt a novel MTT method for multi-AUV based on the fast Labeled Multi-Bernoulli (LMB) filter. In this method, the LMB filter uses belief propagation (BP) to solve the data association problem quickly and effectively approximate the LMB during the update step. And a Gaussian mixture approximation is used to determine the new potential target trajectory based on individual AUV-bearing measurements. Furthermore, we employ the iterator-corrector strategy to perform the fast LMB filter for multi-AUV. The simulation results show that the method performs well in MTT for multi-AUV using bearing-only measurements.